External Governance

External governance is an experimental program within the MO§ES research agenda. It tests a single question: whether an externally supplied structural framework can be recognized, applied, and maintained by an AI system during operation.

The Question

MO§ES investigates whether an externally supplied structural framework can be recognized, applied, and maintained by a system during operation.

The framework is introduced from outside the system. The system did not generate it, and it was not baked in at training time. The question is whether the system can take that framework and operate under it — not merely echo its language, but actually apply it to its own behavior and maintain that application as it works.

Test Flow

The experimental flow introduces a framework and observes whether the system can carry it through its own operations:

external framework introduced
    ↓
system interprets framework
    ↓
system applies it to its own behavior
    ↓
system identifies conflicts
    ↓
system alters / abstains / explains

Each stage is observable. The flow is not assumed to succeed; it is the sequence the experiment checks. Failure at any stage — interpretation, application, conflict identification, or response — is itself a result.

Observable Questions

The program measures behavior against a set of observable questions:

• Can multiple models apply the same framework?
• Do they interpret it consistently?
• Does behavior change?
• Does behavior persist after context changes?
• Does the system merely imitate wording?
• Can it reject malformed or contradictory versions?

These questions are designed to separate genuine application from surface imitation. A system that repeats the framework's vocabulary but does not change its behavior has not applied the framework — it has imitated it. A system that applies the framework under one context but drops it when context shifts has not maintained it.

What This Is Not

This is not a claim that external governance is solved. It is an experimental program testing whether the capability exists and can be measured. The honest framing is investigational: MO§ES is asking the question, building the tests, and reporting what is observed — including negative results.

Relationship to Internal Governance

External governance tests whether a system can operate under a framework it did not generate. This is distinct from two adjacent concepts:

• Self-governance: the system generates its own rules. The framework is internal in origin.
• Alignment: the framework is baked in at training time. The system is shaped to produce desired behavior rather than presented with a framework to apply during operation.

The three differ in where the framework comes from and when it is introduced. External governance is the case where the framework is supplied from outside, at operation time, and the system must take it up itself.

Related Research

• Governance vs Alignment: why governance (execution-time continuity) and alignment (training-time shaping) are complementary, not alternatives
• Persistent Governing State: the mechanism by which a system evaluates its own proposed actions against a persistent reference
• Governance Enforcement: the enforcement mechanisms that make governance operational
• Architecture: how MO§ES operationalizes governance at the execution layer

The Ecosystem

Three properties, one progression. MO§ES™ sits in the middle — the architecture layer.

mos2es.org (commercial) ↓ mos2es.com (architecture) ↓ signalaf.com/wiki (evidence)

Commercial Capability

MO§ES governance capabilities are being piloted for production deployment. For the commercial capability overview, see mos2es.org/pilot. For the evidence layer — definitions, tests, and falsifiers — see signalaf.com/wiki/governance/execution-layer-governance.